Triumph of the City: How Our Greatest Invention Made Us Richer, Smarter, Greener, Healthier and Happier

by Edward Glaeser

  The city-crime connection also reflects the difficulty of law enforcement in big, often anonymous cities. In the game of Clue, players solve a murder by progressively eliminating all the possible suspects. Real cops often do the same thing, but this process is much harder in cities because there are a lot more suspects to consider. As a result, the probability of being arrested for any given crime drops by about 8 percent as city population doubles.

  Crime rates have been reliably correlated with city size, but the differences in crime rates among cities and over time often have little to do with law enforcement, income, or anything else that can be measured. Rio’s slums are famous for their trigger-happy gangs, but Mumbai’s slums are usually quite safe. Despite the prevalence of criminals in the movie Slumdog Millionaire, the overall crime rate in Mumbai is much lower than in urban India as a whole. Mumbai’s slums lack the dangerous feeling I have felt in Rio’s favelas or New York’s poorer areas in the 1970s. This discrepancy isn’t because Mumbai’s police are doing a great job, and Mumbai is poorer than Rio.

  The best explanation for the safety of Mumbai’s slums is that, while these places maybe poor, they’re also well-functioning social spaces, like the Greenwich Village described by Jane Jacobs in her masterpiece The Death and Life of Great American Cities fifty years ago. In these areas, residents watch the streets and alleys. Misbehavior is quickly noticed and dealt with, not by the police, but by the community.

  Within cities too, crime rates move up and down for reasons that are hard to fathom. Murder is the only crime that can be reliably used to measure long-term changes in public safety, because other crimes are often underreported for various reasons. Official crime rates can actually decrease when a police force is particularly inept or corrupt, because people stop reporting most crimes altogether.

  The crime historian Eric Monkkonen assembled data on more than two hundred years of murders in New York. Murder fell from 1800 to 1830 and then rose again, peaking during the Civil War. During the nineteenth century, when street gangs held sway over immigrant neighborhoods and New York’s police were infamously corrupt, there were generally between three and six murders per hundred thousand New Yorkers every year. There does appear to be a weak link between corruption and homicide. Between 1865 and 1961, there were about 12 percent more homicides during years when Tammany Hall was in power than during reform administrations.

  Murders then fell in the late nineteenth century, rose after 1900, and peaked during the Roaring Twenties at 5.4 murders per hundred thousand before falling to a low of 4.1 in the 1950s. The national homicide rate fell by about 29 percent between 1939 and 1959. Between 1960 and 1975, all of the gains that were made during the 1930-1960 period evaporated, and cities became more lawless than ever. The murder rate in New York increased fourfold, to 22 murders per hundred thousand people in 1975.

  Many of the fluctuations in crime rates have no obvious causes. America and New York were getting richer and bigger during all these time periods; poverty or city size can’t explain why crime rose or fell during particular decades. The crime explosion from 1960 and 1975 has been extensively analyzed, but no consensus has emerged. One might guess that this increase can be explained by the rising number of young people during those years (crimes are disproportionately committed by the young), but Steven Levitt estimates that the increase in the number of young people can explain, at most, one fifth of the rise in crime during this period. Other explanations include worsening economic conditions in urban industrial economies or a decline in police effectiveness, but again no measurable variable can really explain the change.

  The inexplicable changes in crime over time and space are, in a sense, the damaging counterpart of the inexplicable explosions in art and creativity that occasionally appear in cities. Both phenomena are examples of the power of social interactions. One artist—Brunelleschi, Haydn—can set off a chain of innovation in his or her city. Likewise, a handful of urban criminals can break down the social norms that keep cities safe, thereby making crime more attractive. The Crips, the vast gang that now supposedly contains more than thirty thousand members, was founded by a few young men. Because cities enhance the influence of individuals—for good and ill—and because the choices and talents of individuals are enormously unpredictable, urban phenomena like crime waves are similarly hard to understand.

  Crime waves may be hard to explain, but their impact can be painfully obvious. Between 1940 and 1960, New York was about as healthy as the rest of the United States. Life expectancies for white men were never more than six months different between New York and the nation as a whole. But between 1960 and 1990, a 2.7-year gap opened up between life expectancy for men in New York and life expectancy for men elsewhere. Men in the country as a whole were getting healthier than those in New York. This gap didn’t appear for women, in part, because the great majority of murder victims are men.

  Plenty of factors contributed to the increasing numbers of dead men in New York. AIDS arrived and started slaying New Yorkers, again mostly men. Death rates from heart disease also increased in New York between the 1960s and 1980s, perhaps because of drug use or stress. New York’s Central Park became a kind of no-man’s-land that only the brave or foolhardy would enter at night. In 1925, the lyricist Lorenz Hart had described the city as a “wond’rous toy just made for a girl and boy.” Fifty years later, the city seemed made for muggers—and was anything but “wond’rous.”

  Then between 1975 and 2005, New York’s murder rate declined from nearly twenty-two to slightly over six deaths per hundred thousand residents. This decline was accompanied by similar drops in the numbers of rapes, robberies, and almost all serious crimes. Just as there is much that was inexplicable about the rise in crime rates, much about the drop in crime reflects social forces beyond measurement or control. John Donohue and Steven Levitt have persuasively argued that the legalization of abortion played some role in the crime drop.

  Moreover, even though crime rates can often change for reasons wholly unrelated to the police, policing does matter. The economics of crime and punishment, pioneered by Gary Becker, starts with the premise that criminals aren’t totally irrational. They respond—just like the rest of us—to incentives. There will be less crime if the expected punishment from crime increases, and the expected punishment depends on the probability of arrest and the severity of punishment after arrest. The rationality of crime actually makes sense of recidivism rates that are often above 90 percent. If criminals are rational and know what to expect from prison before going to jail, then a stint in the slammer shouldn’t change their thinking about their life’s work. No one expects professional basketball players who foul out of a few games to suddenly change their style of play. If crime seemed like a good idea to someone before arrest, why should it look like a bad idea afterward?

  Plenty of statistical work supports the intuitive idea that crime falls as punishments increase, although many studies have found that crime falls more in response to higher arrest rates than in response to longer sentence lengths. The sky-high murder rates in South American cities like Rio and Bogotá can be explained by the low conviction rates for murders. In the United States, about 50 percent of murders lead to a conviction. In Bogotá and Rio, fewer than 10 percent of murderers end up with jail time. It’s not surprising that these places have such extreme crime problems when the costs of committing crime are so low. In Latin America, the more popular response to high crime rates was trying to fix the poverty that accompanies crime. Unfortunately, that strategy has been no more successful there than it was in the United States.

  After America’s cities exploded in crime and rioting in the 1960s, an early consensus argued that we could make our cities safer by making them more prosperous. As long as America focused on solving its poverty problem, the crime problem would automatically be solved. In response to the riots, the Kerner Commission recommended that America “take immediate action to create 2,000,000 new jobs over the next three years—one mil
lion in the public sector and one million in the private sector—to absorb the hard-core unemployed and materially reduce the level of underemployment for all workers, black and white.”

  Unfortunately, no one really knew how to create 2 million new jobs for the urban unemployed, how to solve the poverty problem more generally, or how to stem the decline in urban manufacturing during this era. Moreover, it was far from obvious that rising incomes alone would radically reduce crime. As the 1960s turned into the 1970s, even liberals started arguing for a more direct law-and-order approach to crime prevention.

  In 1973, Nelson Rockefeller, once considered the liberal hope of the Republican Party, signed the Rockefeller drug laws, which mandated a prison sentence of fifteen years to life for anyone possessing four ounces or more of any illegal drug. In the mayoral election of 1977, Ed Koch distinguished himself from his rivals by supporting the death penalty. Koch started a trend, and his successors, including Rudy Giuliani, embraced the “broken windows” theory of policing, which calls for strong penalties for even minor infractions, such as jumping subway turnstiles to avoid paying the fare. Harsher penalties naturally appealed to citizens of a city where criminals seemed to be in control.

  Between 1980 and 2000, the number of inmates in the U.S. criminal system—in prison, in jail, on probation, or on parole—increased from 1.8 million to 6.4 million. Jails don’t reform criminals, but they do stop crime through deterrence and, more important, by keeping criminals off the streets. There is extensive research examining the impact of incarceration on crime levels, and typically when sentence lengths double, crime rates decline by somewhere between 10 and 40 percent. Steven Levitt argues that the incapacitation effect of jails is usually more important than deterrence. One classic study of his used ACLU lawsuits against overcrowding that pushed prisons to let criminals out. After the releases, crime rates rose nearby, and he estimates that as prison populations drop by 10 percent, violent crime increases by 4 percent. Using this estimate, the increase in prison population can explain almost 40 percent of the drop in violent crime during the 1990s.

  Millions of young men have been brought into the prison system for nonviolent drug crimes. Some of these men would have done worse things if they had been free, and their incarceration helped reduce crime rates. But many of them would have led perfectly productive lives. The loss of their freedom and future prospects is the terrible price of reducing crime rates by increasing incarceration rates. I cannot say whether the costs to those prisoners and their communities is outweighed by the benefits of increased public safety, but I fervently hope that we can find less painful means of reducing crime in the future.

  As Becker’s logic suggests, another way to reduce crime levels is just to hire more cops. During the 1990s, the number of police in New York City increased by 45 percent. The number of police increased by 15 percent nationwide. Steven Levitt estimates that as the number of police increases by 10 percent, crime drops by 5 percent. If one accepts that number, then more cops can explain about one seventh of the national drop in crime and perhaps a quarter of New York’s particularly sharp drop in violence. More cops aren’t free, but they do seem to be at least as cost-effective as longer prison stays.

  Is there a free lunch—a way to reduce crime without spending more on police or imprisoning millions of young men? Two strategies have gotten a lot of publicity over the past two decades, both of which aim to improve the flow of information in the police force. One of those strategies uses technology; the other relies on urban interactions. Both seem to be effective, even if we can’t say that with the same confidence as we can about the impact of longer sentences or more police.

  Police forces have long embraced new technologies like fingerprinting, automobiles, lie detectors, two-way radios, and the 911 system. The latest high-tech wave washed over law enforcement in the 1990s in places like New York, where an innovative data-driven system helped target police resources toward troubled areas. The idea seems to have started with a transit cop, Jack Maple, who marked a map of the New York Transit System so that he could see where robberies were most common. He used the map to determine where to assign his cops, following John Snow’s lead in both mapmaking and self-protecting urban innovation.

  Gradually, the system got more sophisticated, and a large contingent of officers would quickly descend on a subway station following a crime there. The number of subway robberies dropped dramatically, and Maple’s idea was borrowed by his new boss, Police Chief William Bratton. Bratton and Maple then created CompStat, a computerized statistical system that allows precinct captains and their superiors to see exactly where crime is taking place and act accordingly. CompStat made the city safer by pinpointing where resources were needed most and making cops more accountable for crimes that occurred on their watch.

  While CompStat relied on snazzy new technology to improve enforcement, “community policing” relied on personal contact. At its heart, community policing simply means that police should maintain good relations with a neighborhood and use face-to-face interactions to glean information that can help prevent crimes. Criminals, especially those in gangs, are often protected by their neighborhoods, both out of fear and because even the deadliest gangs often take care of their neighbors. But while the idea of community policing may be simple, the execution can be quite difficult.

  Cops are often outsiders because they come from a different place or race than those in the communities they patrol. Also, earlier attempts at police professionalization had often worked to break the link between cop and community. Many cities adopted police rotation—periodically moving cops to new neighborhoods—because they thought it would reduce corruption by weakening the links between crooked cops like “Clubber” Williams and residents who would pay them off. But the riots of the 1960s, which often started with local groups attacking cops, pushed police departments to invest more in community relations in an attempt to improve their relationship with hostile neighborhoods.

  In 1992, violence came to Boston’s Morning Star Baptist Church when a funeral became a fight between rival gangs. A coalition of community leaders, drawn mostly from the clergy, came together and created the Ten Point Coalition, a league of religious leaders that is working to reduce violence in the city’s poorer neighborhoods. The support of these leaders gave Boston’s police better ties to troubled areas and made them considerably more effective. Crime rates fell dramatically.

  Today, the Boston Police Department has a number of community policing initiatives, including “safe streets teams” and a bevy of neighborhood advisory councils. The officers who work these beats say that women are often their key contacts, and minority, female police officers can be particularly effective in building these bridges.

  While it’s hard to see the downside in policing strategies that leverage the urban ability to spread knowledge across space, there’s little hard data to support the view that either community policing or CompStat-like programs have driven down crime rates dramatically. The introductions of these methods are rarely controlled experiments. But numerous case studies do suggest that they can help make city streets safe.

  Neither CompStat nor community policing could protect the 2,794 New Yorkers who died when two Boeing 767 airplanes destroyed the World Trade Center on September 11, 2001. Despite the courage of so many New Yorkers, like the dead hero whose picture still shines from the firehouse on my mother’s street, many doubted the city’s ability to recover. They feared that urban concentration would present an irresistible target to terrorists who wanted to strike at the very core of our civilization. But there is little evidence to suggest that cities can’t survive the threat of terrorism. Across countries, historically, terrorism does not deter either urbanization or the construction of tall buildings. Both Jerusalem and London have faced ongoing terrorist activity, but neither city’s population growth seems to have declined as a result. Cities have powerful resources—large police forces, observant citizens, strong infrastructure—that h
ave so far enabled them to protect themselves against even the most terrifying threats.

  Health Benefits

  A mass of people living on a small amount of land creates enormous health risks, but as of 2007, a child born in New York City could, if current death rates continue, expect to live one-and-a-half years longer than a child born in the United States as a whole. Los Angeles, Boston, Minneapolis, San Francisco, and many other cities can also boast age-adjusted death rates lower than the national average. The average life expectancy in counties with more than five hundred people per square mile is nine months longer than in counties with fewer than a hundred people per square mile. Between 1980 and 2000, life expectancies increased six months more in counties with more than five hundred people per square mile than in counties with less than that density.

  New Yorkers’ good health didn’t just happen. It took massive public investment to bring in potable water. It took a tough, quasi-military leader who drastically increased his department’s expenses to make Manhattan’s streets clean. Lots of cops and higher rates of imprisonment made New York City safe. Every battle was won by accountable and empowered public leaders who spent huge amounts of money and enlarged the public sector. The troubled cities of the developing world must go through a similarly difficult process if they are to become safe and clean.

  But these investments can only explain why large cities are no longer killing fields. The decline in urban infectious diseases and homicides can’t explain why many cities like New York are healthier than the nation as a whole. It is easiest to understand why the death rate for Manhattanites between the ages of twenty-five and thirty-four is 60 percent lower than the comparable rate for the country. Accidents and suicides are the two leading causes of death for these younger people, and these are both rarer in big cities. New Yorkers in this age group are more than 75 percent less likely to die in a motor vehicle accident than their counterparts nationwide. Driving drunk is far more deadly than taking the bus while intoxicated.